Speaking Engagements & Private Workshops - Get Dean Bubley to present or chair your event

Need an experienced, provocative & influential telecoms keynote speaker, moderator/chair or workshop facilitator?
To discuss Dean Bubley's appearance at a specific event, contact information AT disruptive-analysis DOT com

Monday, October 26, 2009

HSPA in 2.6GHz?

I'm wondering.....

... if LTE looks like it might be delayed, for example because of poorer-than-expected performance, difficult optimisation, continued wrangling over voice/SMS implementation, or because operators don't want to be strong-armed into IMS...

... then does it start to make sense to put HSPA/HSPA+ into the 2.6GHz bands, especially given the flurry of upcoming auctions in 2010/2011?

After all, HSPA is a "known quantity" in terms of radio deployment and operation, it's not too difficult to add another band to existing handset platforms, and it's got voice built-in out of the box.

Let's imagine a situation in markets with existing consumer use of mobile broadband, say Europe or Australia or parts of Asia. Now imagine the end of 2012 - there's a lot of 2.6GHz spectrum that's now owned by MNOs. LTE still has teething problems for whatever reason... and in any case, there's several hundred million PCs, dongles, smartphones and other gizmos running on HSPA, albeit only on existing bands like 2.1GHz. I've got to believe that a 2.1/2.6GHz HSPA+ netbook on sale for Xmas 2012 is going to be cheaper and more reliable than a 2.1GHz HSPA + 2.6GHz LTE one - and with broadly similar performance and network efficiency.

On the same theme, do any readers familiar with the innards of UMTS specifications think it might be possible to tweak R9 or R10 HSPA to support flexible channel size, from 5MHz-only to something more like LTE's range of options.


Anonymous said...

I believe 3GPP is looking at HSPA in 20 MHz i.e., 4 carriers of 5 MHz bundled together..so may be in R10 time frame it might be possible.

Johan Nilsson said...

Changing the bandwidth is only possible by bundling 5 MHz carriers. (They are actually not 5 MHz but slightly less, so you could possibly squeeze in two in maybe 9 MHz or so, but there is no way that you will have a 2.5 MHz or 1.4 MHz WCDMA carrier.
The multi carrier transmission seems to be a pretty straight forward solution. (Actually this was suggested in the early 3G days by Qualcomm to bundle three cdma2000 carriers in 5 MHz, but that was not the 3GPP way then ;-)

Gabe said...

Hi Dean, My opinion: it doesn't make sense. Operators will deploy new technology in new spectrum. Fwiw, LTE appears to be performing at least as well as the early simulations predicted.

Sami said...

I work for Nokia, but the opinions below are mine.

To begin with, I'm not sure operators need to be strong-armed into IMS to offer LTE connectivity for PCs, dongles and other gizmos.

Network that is opened in time for Xmas market in 2012, needs to serve consumers over several years - not one, two or three. Since any deployment effectively hinders introduction of another tech on the same band, majority of operators would rather deploy as future proof tech as possible on new bands.

To add another band to existing handset platforms is for sure easier than development of new tech, but if you assume broadly similar performance and network efficiency as with LTE, then HSPA is not the one we have today or tomorrow. And as you know, not all features in future HSPA releases are SW only upgrades even on network side, meaning it is far from being free (or easy).

Yes, tweaks are prepared in 3GPP for HSPA to support flexible channel size, from 5MHz-only to wider bandwidths, but not for narrower. How easy it is to scale HSPA transceivers up to 20MHz bands, depends on the final specifications, but for sure there will be some additional cost.

Overall, I think it is unlikely that HSPA on 2.6GHz would become main stream technology.

Arun Demeure said...

I agree with the previous commenters; I don't think 2.6GHz has much of a chance. Part of the problem is that very few RF chips will support it for 3G and these take quite some time to design; by the time everyone realizes LTE is in trouble in your scenario, it would be too late to get those out of the door.

As Martin on mobilesociety.typepad.com noted a few times, there's still quite a lot of unused spectrum nearly everywhere. For the few places where operators are out, it seems to be that aggressive picocell deployments would be a better solution than adding yet another carrier.

I think the end-game for HSPA+ is 84Mbps (64QAM, MIMO, 2x5MHz) even in the absolute worst case for LTE (in the best case, anything above 21Mbps will remain niche). Of course, that would be very very competitive with all early LTE deployments at 50-100Mbps.

Dean Bubley said...

Thanks for the feedback.

Bundled HSPA sounds interesting - I'll watch out for it.

Gabe - yes, that's certainly the received wisdom that "New Spectrum = Latest Technology", but I'm questioning it here. It's not obvious to me that LTE offers that much extra beyond HSPA. Headline peak speeds are irrelevant most of the time - it's average performance that's important.

This is especially true as 3GPP has missed a trick by specifically looking to enabling new business models in LTE. It could have looked at SIM-free modes of use, for example.

Sami - fair points, although it's not obvious to me that LTE investment can show a return solely on dongles & embedded devices; it will need to drive some voice & SMS revenues too.

Arun - yes, good points. But this post was partly intended to make RF vendors think again about supporting 2.6GHz HSPA, as I think it represents a possible opportunity given the scenario I have outlined.